hanlint 0.0.1

package/src/rules/sentence/nounPile.js

@@ -0,0 +1,20 @@
+// @ts-check
+import { SENTENCE, finding } from "../finding.js";
+
+export const name = "nounPile";
+
+/**
+ * @param {import("../../fingerprint/build.js").DocumentPrint} doc
+ * @param {import("../../config/settings.js").Config} config
+ */
+export function run(doc, config) {
+  const findings = [];
+  for (const sentence of doc.sentences) {
+    if (sentence.nounRun >= config.nounPileMin) {
+      findings.push(
+        finding(name, sentence.line, sentence.text, `명사 ${sentence.nounRun}개가 조사 없이 이어진다. 관계가 표시되지 않아 독자가 조사를 끼워 넣는다. 동사로 되돌린다`, null, "error", SENTENCE, sentence.index),
+      );
+    }
+  }
+  return findings;
+}

package/src/rules/sentence/redundantPair.js

@@ -0,0 +1,9 @@
+// @ts-check
+import { dictionaryFindings } from "../shared/dictionaryRule.js";
+
+export const name = "redundantPair";
+
+/** @param {import("../../fingerprint/build.js").DocumentPrint} doc */
+export function run(doc) {
+  return dictionaryFindings(doc, "redundantPair", name);
+}

package/src/rules/sentence/translationese.js

@@ -0,0 +1,9 @@
+// @ts-check
+import { dictionaryFindings } from "../shared/dictionaryRule.js";
+
+export const name = "translationese";
+
+/** @param {import("../../fingerprint/build.js").DocumentPrint} doc */
+export function run(doc) {
+  return dictionaryFindings(doc, "translationese", name);
+}

package/src/rules/shared/codeBlocks.js

@@ -0,0 +1,44 @@
+// @ts-check
+/**
+ * 코드 블록을 언어와 본문으로 나눈다. code 부류 규칙이 함께 쓴다. 파이썬 rules/shared/codeBlocks.py 와 같다.
+ * 펜스 첫 줄의 언어 표기 (```python) 를 읽고 본문 줄에 원문 줄 번호를 붙인다. text 펜스는 출력이다.
+ */
+import { CODE } from "../../document/model.js";
+
+const FENCE_LANGUAGE = /^\s*(?:```|~~~)\s*([A-Za-z0-9_+-]*)/;
+const CLOSING_FENCE = /^\s*(?:```|~~~)\s*$/;
+
+/**
+ * @typedef {object} CodeBlock
+ * @property {number} index
+ * @property {number} startLine
+ * @property {string} language
+ * @property {[number, string][]} lines (원문 줄 번호, 코드 줄)
+ * @property {string} text
+ * @property {boolean} isOutput
+ */
+
+/** @param {import("../../fingerprint/build.js").DocumentPrint} doc @returns {CodeBlock[]} */
+export function codeBlocksOf(doc) {
+  /** @type {CodeBlock[]} */
+  const blocks = [];
+  for (const block of doc.blocks) {
+    if (block.kind !== CODE) continue;
+    const raw = block.text.split("\n");
+    const match = FENCE_LANGUAGE.exec(raw[0]);
+    const language = (match ? match[1] : "").toLowerCase();
+    let body = raw.slice(1);
+    if (body.length && CLOSING_FENCE.test(body[body.length - 1])) body = body.slice(0, -1);
+    /** @type {[number, string][]} */
+    const lines = body.map((line, offset) => [block.startLine + 1 + offset, line]);
+    blocks.push({
+      index: block.index,
+      startLine: block.startLine,
+      language,
+      lines,
+      text: lines.map(([, line]) => line).join("\n"),
+      isOutput: ["text", "", "output", "console"].includes(language),
+    });
+  }
+  return blocks;
+}

package/src/rules/shared/dictionaryRule.js

@@ -0,0 +1,35 @@
+// @ts-check
+/** 사전 규칙의 공통 구현. cliche, translationese, redundantPair, japaneseLoan 이 사전 이름만 바꿔 쓴다. */
+import { ERROR, SENTENCE, finding } from "../finding.js";
+
+/**
+ * @param {import("../../fingerprint/build.js").DocumentPrint} doc
+ * @param {string} dictionary
+ * @param {string} ruleName
+ * @param {string} [severity]
+ * @returns {import("../finding.js").Finding[]}
+ */
+export function dictionaryFindings(doc, dictionary, ruleName, severity = ERROR) {
+  const findings = [];
+  for (const sentence of doc.sentences) {
+    for (const match of sentence.matches) {
+      if (match.dictionary !== dictionary) continue;
+      const fix = match.fix !== null ? sentence.text.slice(0, match.start) + match.fix + sentence.text.slice(match.end) : null;
+      findings.push(
+        finding(
+          ruleName,
+          sentence.line,
+          sentence.text,
+          `\`${match.text}\` ${match.why} (${match.source})`,
+          fix,
+          severity,
+          SENTENCE,
+          sentence.index,
+          match.fix !== null ? match.text : null,
+          match.fix,
+        ),
+      );
+    }
+  }
+  return findings;
+}

package/src/rules/shared/runs.js

@@ -0,0 +1,22 @@
+// @ts-check
+/** 연속 구간 찾기. 같은 값이 이어지는 자리를 [시작 위치, 길이, 값] 으로 준다. */
+
+/**
+ * @template T
+ * @param {T[]} values
+ * @param {number} minLength
+ * @returns {[number, number, T][]}
+ */
+export function runsOf(values, minLength) {
+  /** @type {[number, number, T][]} */
+  const runs = [];
+  let start = 0;
+  while (start < values.length) {
+    let end = start;
+    while (end + 1 < values.length && values[end + 1] === values[start]) end += 1;
+    const length = end - start + 1;
+    if (length >= minLength) runs.push([start, length, values[start]]);
+    start = end + 1;
+  }
+  return runs;
+}

package/src/rules/structure/emojiBullet.js

@@ -0,0 +1,34 @@
+// @ts-check
+import { LIST } from "../../document/model.js";
+import { DOCUMENT, finding } from "../finding.js";
+
+export const name = "emojiBullet";
+// 이모지 범위. 코드포인트로 만든다 (도구가 이스케이프를 건드리지 않게). 아스트랄 범위라 u 플래그가 필요하다.
+const EMOJI =
+  "[" +
+  String.fromCodePoint(0x1f300) +
+  "-" +
+  String.fromCodePoint(0x1faff) +
+  String.fromCodePoint(0x2600) +
+  "-" +
+  String.fromCodePoint(0x27bf) +
+  String.fromCodePoint(0x2b50) +
+  String.fromCodePoint(0x2b06) +
+  "]";
+const BULLET_WITH_EMOJI = new RegExp(/^\s*(?:[-*+]|\d+[.)])\s+/.source + EMOJI, "u");
+
+/** @param {import("../../fingerprint/build.js").DocumentPrint} doc */
+export function run(doc) {
+  const findings = [];
+  for (const block of doc.blocks) {
+    if (block.kind !== LIST) continue;
+    const lines = block.text.split("\n");
+    for (let offset = 0; offset < lines.length; offset++) {
+      if (BULLET_WITH_EMOJI.test(lines[offset])) {
+        findings.push(finding(name, block.startLine + offset, lines[offset].trim(), "목록 항목이 이모지로 시작한다. 이모지를 지우고 첫 낱말이 내용을 말하게 한다", null, "error", DOCUMENT, block.index));
+        break;
+      }
+    }
+  }
+  return findings;
+}

package/src/rules/structure/headingSentence.js

@@ -0,0 +1,17 @@
+// @ts-check
+import { DOCUMENT, finding } from "../finding.js";
+
+export const name = "headingSentence";
+const SENTENCE_ENDINGS = ["니다", "세요", "십시오", "합시다", "."];
+
+/** @param {import("../../fingerprint/build.js").DocumentPrint} doc */
+export function run(doc) {
+  const findings = [];
+  for (const [, text, line] of doc.headings) {
+    const trimmed = text.trimEnd();
+    if (SENTENCE_ENDINGS.some((ending) => trimmed.endsWith(ending))) {
+      findings.push(finding(name, line, text, "절 제목이 본문 문장을 복사했다. 대상이나 질문을 짧게 쓴다", null, "error", DOCUMENT, -1));
+    }
+  }
+  return findings;
+}

package/src/rules/structure/headingSkip.js

@@ -0,0 +1,18 @@
+// @ts-check
+import { DOCUMENT, finding } from "../finding.js";
+
+export const name = "headingSkip";
+
+/** @param {import("../../fingerprint/build.js").DocumentPrint} doc */
+export function run(doc) {
+  const findings = [];
+  /** @type {number | null} */
+  let previous = doc.frontmatter.title ? 1 : null;
+  for (const [level, text, line] of doc.headings) {
+    if (previous !== null && level > previous + 1) {
+      findings.push(finding(name, line, text, `H${previous} 다음에 H${level} 이 온다. 한 수준씩 내려간다`, null, "error", DOCUMENT, -1));
+    }
+    previous = level;
+  }
+  return findings;
+}

package/src/rules/structure/headingUniform.js

@@ -0,0 +1,36 @@
+// @ts-check
+import { DOCUMENT, finding } from "../finding.js";
+
+export const name = "headingUniform";
+
+/**
+ * @param {import("../../fingerprint/build.js").DocumentPrint} doc
+ * @param {import("../../config/settings.js").Config} config
+ */
+export function run(doc, config) {
+  const headings = doc.headings.filter(([level]) => level === 2);
+  if (headings.length < 3) return [];
+  // 파이썬 Counter.most_common 은 같은 수면 먼저 들어온 것이 이긴다. 삽입 순서를 지킨다.
+  /** @type {Map<string, number>} */
+  const lastChars = new Map();
+  for (const [, text] of headings) {
+    if (!text.trim()) continue;
+    const trimmed = text.trimEnd();
+    const char = trimmed[trimmed.length - 1];
+    lastChars.set(char, (lastChars.get(char) ?? 0) + 1);
+  }
+  let best = "";
+  let count = -1;
+  for (const [char, n] of lastChars) {
+    if (n > count) {
+      best = char;
+      count = n;
+    }
+  }
+  if (count / headings.length >= config.headingUniformRatio) {
+    return [
+      finding(name, headings[0][2], headings.map(([, text]) => text).join(" / "), `H2 ${headings.length}개 중 ${count}개가 \`${best}\` 로 끝난다. 목차가 한 어미로 끝나면 과정이 아니라 나열로 읽힌다. 형태를 섞는다`, null, "error", DOCUMENT, -1),
+    ];
+  }
+  return [];
+}

package/src/rules/structure/introLong.js

@@ -0,0 +1,20 @@
+// @ts-check
+import { PARAGRAPH, finding } from "../finding.js";
+
+export const name = "introLong";
+
+/**
+ * @param {import("../../fingerprint/build.js").DocumentPrint} doc
+ * @param {import("../../config/settings.js").Config} config
+ */
+export function run(doc, config) {
+  if (!doc.bodySections.length) return [];
+  const paragraphs = doc.intro.paragraphs;
+  if (paragraphs.length > config.introMaxParagraphs) {
+    const over = paragraphs[config.introMaxParagraphs];
+    return [
+      finding(name, over.startLine, over.sentences.length ? over.sentences[0].text : "", `도입 산문 문단이 ${paragraphs.length}개다. ${config.introMaxParagraphs}개를 넘지 않는다. 첫 코드 블록이 그 안에 온다`, null, "error", PARAGRAPH, over.index),
+    ];
+  }
+  return [];
+}

package/src/rules/structure/sectionNoProse.js

@@ -0,0 +1,16 @@
+// @ts-check
+import { LIST, PROSE } from "../../document/model.js";
+import { SECTION, finding } from "../finding.js";
+
+export const name = "sectionNoProse";
+
+/** @param {import("../../fingerprint/build.js").DocumentPrint} doc */
+export function run(doc) {
+  const findings = [];
+  for (const section of doc.bodySections) {
+    const kinds = new Set(section.blockKinds);
+    if (!kinds.size || kinds.has(PROSE) || kinds.has(LIST)) continue;
+    findings.push(finding(name, section.startLine, section.title, "설명글 없이 코드, 표, 이미지만 있는 절이다. 절은 제목과 시각 자료와 설명글로 짠다", null, "error", SECTION, section.index));
+  }
+  return findings;
+}

package/src/rules/structure/sectionResult.js

@@ -0,0 +1,37 @@
+// @ts-check
+import { CODE, IMAGE, TABLE } from "../../document/model.js";
+import { NOTICE, SECTION, finding } from "../finding.js";
+
+export const name = "sectionResult";
+const RESULT_KINDS = [CODE, IMAGE, TABLE];
+const FILE_NAME = /[\w가-힣-]+\.(?:csv|xlsx|xls|parquet|json|png|jpg|svg|db|sqlite|txt|py|pdf|html|md)\b/;
+const OUTPUT_WORDS = ["출력", "화면에", "찍힙니다", "나옵니다", "뜹니다", "보입니다", "만들어집니다", "생깁니다", "저장됩니다"];
+
+/** @param {import("../../fingerprint/build.js").SectionPrint} section */
+function leavesResult(section) {
+  if (section.blockKinds.some((kind) => RESULT_KINDS.includes(kind))) return true;
+  for (const paragraph of section.paragraphs) {
+    for (const sentence of paragraph.sentences) {
+      if (FILE_NAME.test(sentence.text) || OUTPUT_WORDS.some((word) => sentence.text.includes(word))) return true;
+    }
+  }
+  return false;
+}
+
+/**
+ * @param {import("../../fingerprint/build.js").DocumentPrint} doc
+ * @param {import("../../config/settings.js").Config} config
+ */
+export function run(doc, config) {
+  const findings = [];
+  const body = doc.bodySections;
+  body.forEach((section, index) => {
+    if (index === body.length - 1) return;
+    if (section.paragraphs.length <= config.sectionResultMinParagraphs) return;
+    if (leavesResult(section)) return;
+    findings.push(
+      finding(name, section.startLine, section.title, "이 절에 독자가 확인할 결과 (코드, 표, 그림, 파일, 출력) 가 없다. 만들거나 확인할 것을 하나 넣는다", null, NOTICE, SECTION, section.index),
+    );
+  });
+  return findings;
+}

package/src/text.js

@@ -0,0 +1,63 @@
+// @ts-check
+/** 파이썬 str 메서드와 같은 뜻의 문자열 도우미. 두 구현의 결과가 글자 단위로 같아야 해서 따로 둔다. */
+
+/**
+ * 파이썬 str.splitlines(). 끝의 빈 줄을 만들지 않는다.
+ * @param {string} text
+ */
+export function splitLines(text) {
+  if (text === "") return [];
+  const lines = text.split(/\r\n|\r|\n/);
+  if (lines[lines.length - 1] === "") lines.pop();
+  return lines;
+}
+
+/** 파이썬 str.split() (인자 없음). 공백 덩어리로 나누고 빈 것은 버린다. @param {string} text */
+export function splitWords(text) {
+  return text.split(/\s+/).filter((word) => word.length > 0);
+}
+
+/** 파이썬 str.strip(chars). @param {string} text @param {string} chars */
+export function stripChars(text, chars) {
+  let start = 0;
+  let end = text.length;
+  while (start < end && chars.includes(text[start])) start++;
+  while (end > start && chars.includes(text[end - 1])) end--;
+  return text.slice(start, end);
+}
+
+/** 파이썬 str.rstrip(chars). @param {string} text @param {string} chars */
+export function rstripChars(text, chars) {
+  let end = text.length;
+  while (end > 0 && chars.includes(text[end - 1])) end--;
+  return text.slice(0, end);
+}
+
+/** 어절 수. 파이썬 len(text.split()). @param {string} text */
+export function wordCount(text) {
+  return splitWords(text).length;
+}
+
+/** @param {number[]} values */
+export function mean(values) {
+  return values.reduce((a, b) => a + b, 0) / values.length;
+}
+
+/** 모집단 표준편차 (파이썬 statistics.pstdev). @param {number[]} values */
+export function pstdev(values) {
+  const m = mean(values);
+  return Math.sqrt(values.reduce((sum, v) => sum + (v - m) ** 2, 0) / values.length);
+}
+
+/** 부분 문자열 개수 (파이썬 str.count). @param {string} text @param {string} needle @param {number} [end] */
+export function countIn(text, needle, end = text.length) {
+  let count = 0;
+  let from = 0;
+  const limit = end;
+  while (true) {
+    const at = text.indexOf(needle, from);
+    if (at < 0 || at + needle.length > limit) return count;
+    count++;
+    from = at + needle.length;
+  }
+}